I think 1070 supports RTX but in software driver mode, they do not have the actual RTX cores if I am not mistaken. You could try CPU+GPU hybrid in CUDA and OPTIX if you a CPU witth many cores.
1070 gets around render score of 600 on Opendata, while something like 3090 gets 6000 and 4090 gets around 12000.
The highest CPU score on Opendata comes from the 2X AMD EPYC 9654 96-Core Processor at 3500 score. You can see how slow even a 96 cores beasty CPU is in Cycles compared to something like an RTX 3080.
Most CPUs will be just a bottleneck for a true RTX card with OPTIX rendering.
In addition to kkar’s answer, the main reason you also don’t see multiple render squares, is that the default tile setting is now much higher, at 2048 compared to old settings at like 64 or 128.
So unless you are outputting a large image, like at 4K, then it’s just doing the whole image in one go.
Leave CPU off (unless you really have a massive core CPU, like a 64 core threadripper) and set the GTX 1070 to CUDA.
OptiX will still work, but as the tests in my video showed, it will be slower then using CUDA for a non-RTX card.
Thanks for that!
Yeah i did many test while back with all the modes. Hybrid was indeed kind of a waste. The GPU always was waiting for the cpu to finish. Unless the GPY can change its cube size at the end, its always slower in most cases
Thanks for the answer. I did try smaller cubes. 64x 128x 256x, 512x. All showed just a single cube when using Hybrid mode in BLender 3.5 with amd Ryzen 7 4800h and GTX 1660 ti
So out of interest and to keep my knowledge upto date, I just tried it myself, only using the CPU and yeah, smaller tile sizes still only result in a single ‘tile’ being rendered, unlike older version of Blender.
It was still using all cores at 100% to render that tile, but even so, only doing one.
I tired with just the GPU on a small tile size and it was the same, only one ‘tile’ being rendered, while overall being way slower compared to just letting the GPU render the whole image in one hit.
So at this stage, the general take away is, don’t use the CPU unless you run out of VRAM. Don’t use small tile sizes (likely explains why the default is 2048) and if you can manage it, get a RTX based card and use OptiX for the fastest render possible.
While OpenCL is missing right now we are actively working with AMD and Intel on alternatives. It’s just a matter of time until support for these GPUs comes back. The precise APIs we will use is not something that is set in stone, but note that there are open standards like SYCL that are more modern alternatives to OpenCL.
For macOS/Metal, we think the new architecture will be easier to port and provide better performance. There’s nothing specific we can announce regarding that, but it’s probably just a matter of time until this happens.
There are lots of good feature suggestions in the comments. There are of course many improvements we plan to do, this blog post is specifically about the Cycles X research project, it’s not a complete overview of what will happen this year or next.
Regarding caustics, path guiding algorithms mentioned in the post are one way of rendering those more efficiently.
Tiled rendering will be back to support high res renders, but in a different form. All devices and CPU cores will likely be cooperating on the same tile.
Which means of course if one makes the tile too small it would be massively inefficient and even worse with hybrid rendering as the CPU cores get in the way of the GPU.
Right, I think that is why it is really not so useful to have the CPU on in OPTIX. It is literally waste of electricity. No consumer CPU will get to the level of RTX GPU rendering with 3070+ or 4070+ in terms of speed and efficiency in Cycles. The CPU only rendering in Cycles is mostly a renderfarm thing at this point. However the hybrid rendering could still be useful if you own a non RTX card and a CPU with many cores.
I have a 12900k (quite a recent CPU with a lot of cores) and a 3080TI, take a look at the the Opendata render scores, not even close.
Intel Core i9-12900K 390.35
NVIDIA GeForce RTX 3080 Ti 5957.81